CRITERIOS DE EVALUACIÓN

The majority of early type galaxies, i.e. ellipticals and lenticulars, in clusters lie along a linear relation in a color-magnitude diagram (hereafter CMD). This relation is often referred to as the red sequence. The red sequence is a narrow and almost horizontal line, i.e. it consists of ellipticals with varying magnitudes, yet nearly constant col- ors (Visvanathan & Sandage, 1977; Annis et al., 1999). It shows a slight negative slope, which implies that fainter early type galaxies possess somewhat bluer colors than their more luminous counterparts (Baum,1959;Visvanathan & Sandage,1977;

Lugger,1984;Metcalfe et al.,1994). It was also shown that this slope evolves with redshift (Gladders et al.,1998), and furthermore varies with the richness of the clusters (Stoughton et al.,1998). The overall color of the red sequence is redder for more dis- tant structures, and consequently can be used for an approximate redshift estimation of clusters (Gladders & Yee,2000).

According to Aragon-Salamanca et al. (1993), Rakos & Schombert (1995), van Dokkum & Franx(1996),Kelson et al.(1997),Ellis et al.(1997),Bender et al.(1998),

van Dokkum et al. (1998),Stanford et al. (1998),Kodama et al. (1998), Barrientos

(1999), andJørgensen et al.(1999) among others, the majority of the stellar popula- tion of the cluster ellipticals seems to have formed at high redshifts of z>2, and the galaxies then appear to have followed a passive evolution leading to the stringent color and magnitude correlation, that can be observed for all cluster types and redshifts:

• Bower et al.(1992a,b) examined the CMDs of two local structures, the rich Coma cluster, as well as the comparatively poor Virgo cluster. They were able to show that both structures exhibit distinct red sequences with only small scatters. • The low redshift cluster survey ofLopez-Cruz & Yee(2000) consists of 45 X-ray selected Abell clusters of various Abell richnesses, Bautz-Morgan and Rood- Sastry classes up to a redshift of z<0.2. The authors demonstrated that every cluster has a red sequence and that the sequences seem to be very homogeneous from one cluster to the next.

• Smail et al.(1998) used ten optically selected X-ray luminous clusters with red- shifts ranging from 0.22<z<0.28, andBarrientos(1999) examined eight op- tically selected clusters in the redshift range 0.39<z<0.48. In both studies the clusters have red sequences with very homogeneous scatters and colors. • Oke et al.(1998) examined a sample of nine optically selected clusters located at

redshifts 0.6<z<0.9. Two of these structures were spectroscopically verified as clusters and had detailed information. Both objects show conspicuous red sequences (Oke et al.,1998;Stanford et al.,1998;Gladders et al.,1998).

68 CHAPTER 5. AUTHENTICITY OF STRUCTURES

• The existence and behavior of red sequences for clusters with redshifts above unity is not well studied yet, simply because of the low number of corresponding candidates. However red sequences were found for a number of high redshift structures. E.g.da Costa et al.(1999) showed red sequences in two clusters at redshifts of approximately unity.Ben´ıtez et al.(1999) proved the existence of a red sequence in a cluster at z=1.01. Further examples of a strong red sequence were found in clusters at redshifts z=1.206 (Dickinson,1995), and z=1.273 (Stanford et al.,1997).

5.4.2 Application to structure finding

As described in Sect.1.1, galaxy clusters are dominated by early type galaxies, and their fraction of ellipticals can be up to 35%. Due to the correlation between the overall color of the red sequence and the redshift of the cluster, CMDs can be used to approximately deproject two-dimensional multi-color surveys.

This characteristic has been used for cluster finding by various authors:

• The cluster red sequence method (hereafter CRS;Gladders & Yee 2000, andYee & Gladders 2002) was designed to find structures in the Red Sequence Cluster Survey (RCS;Gladders 2002), a two-band imaging survey in the filters R and z0. The CRS algorithm makes use of the projected positions, the R−z0 color, its error, and the z0 apparent magnitude. Clusters are detected as overdensities in the color and magnitude space, as well as in projected positions: A series of overlapping color-magnitude slices removes fore- and background galaxies. Next, the algorithm searches for peaks in the weighted surface density of the individual subsets.

• The Voronoi tessellation technique ofKim et al. (2002) (see Sect. 5.2.2) also uses a binning in color-magnitude space to deproject the dataset. Their method was developed for the multi-band imaging data of the SDSS, with observations in the u∗,g∗,r∗,i∗,and z∗filters. A series of overlapping color and magnitude cuts was used in the g∗−r∗ vs. r∗space and Voronoi tessellations were run on the resulting galaxy subsets.

• Another algorithm was designed to detect structures in the multi-band SDSS imaging data, the cut-and-enhance (hereafter CE) method ofGoto et al.(2002). Its use of color and magnitude information is more refined than theKim et al.

(2002) approach. In contrast to the latter, the CE uses ten overlapping color- magnitude cuts in three different color spaces: r∗−i∗vs. r∗, i∗−z∗vs. r∗, and g∗−r∗vs. r∗. It furthermore utilizes four overlapping color-color cuts in r∗−i∗ vs. g∗−r∗, and i∗−z∗vs. r∗−i∗. All of those 34 cuts are done independently of each other. A density enhancement algorithm, that amplifies galaxy pairs, that are close in color, as well as angular separation, is then applied to the individual subsets. In the last step, the 34 cluster candidate lists are merged into a final cluster catalog.

5.4. COLOR-MAGNITUDE DIAGRAMS 69

Using red sequences in color-magnitude diagrams as tracers of galaxy clusters has a number of advantages. As was shown in Sect.5.4.1, red sequences seem to exist independently of cluster richness, shape, radial profile, or luminosity function. Fur- thermore, the assumptions on colors of cluster members appear to be very robust and relatively independent of cluster type. Consequently, the use of red sequence sensitive methods should yield structure catalogs, that are rather unbiased with respect to any cluster model assumptions. The connection between the red sequence overall color and redshift, on the other hand, turns the red sequence into an interesting tool for the deprojection of elliptical galaxy populations.

However, in case of very sparsely sampled structures, i.e. structures with only few members brighter than the limiting magnitude, the red sequence can become very dif- ficult to detect. This is simply due to the fact, that the sequence usually stands out as an overdensity in color above the fore- and background distribution of other galaxies, while in the case of few visible members, the red sequence signal is dwarfed by the fore- and background noise.

In order to test the authenticity of the MUNICS EXT-FOF clusters, m_V−m_Ivs. m_I CMDs are made for every structure candidate.

First, the colors and magnitudes of the EXT-FOF members of a given cluster, that are neither V , nor I band drop-outs are plotted and highlighted. Then, a subset of ob- jects is selected from the MUNICS galaxy catalog. This subset consists of all galaxies i within a projected aperture of 5rcaround the center of the EXT-FOF cluster candidate, independent of their own photometric redshift

2 sinθic

2 dA H0,ΩM,ΩΛ,z

≤5rc, (5.26)

whereθ_icis the angle between galaxy i and the cluster center, and z is the mean redshift

of the EXT-FOF cluster. The core radius is again set to rc=0.125 h−1Mpc. All of these galaxies, that are neither V , nor I band drop-outs, are then added to the plot, to test whether the EXT-FOF algorithm ignores a significant number of possible cluster candidates.

Finally, a model red sequence for the structure is plotted. The model color- magnitude tracks are taken fromGladders & Yee (2000), and are based on publica- tions byColeman et al.(1980), andKodama(1997). Their models are given in AB magnitudes for ten redshifts ranging from 0.1 to 1.0, and for an open cosmology with H₀=70 km s−1Mpc−1,Ω_M=0.2, andΩ_Λ=0. These models are transformed into the Vega magnitude system used for MUNICS, and are also corrected for a flat cosmol- ogy with H₀=70 km s−1Mpc−1,Ω_M=0.3, andΩ_Λ=0.7 (see Sect.6.2.1). Fig.5.3

illustrates the resulting expected variation of the cluster red sequences with redshift in the m_V−m_Ivs. m_I color-magnitude diagrams. In order to compare the distribution of EXT-FOF structure members with the red sequence models, the tracks are interpolated and plotted for the structure mean redshift z.

70 CHAPTER 5. AUTHENTICITY OF STRUCTURES

Figure 5.3: m_V−m_I vs. m_Imodel red sequences. The numbers above the tracks show the redshift.

In case of realistic structures, the majority of the EXT-FOF members are expected to lie on an almost horizontal line. The model track at z is not necessarily the best- fitting sequence, simply because of the large uncertainties in the photometric redshifts and the corresponding uncertainty in the cluster mean redshift. If the number of EXT- FOF members is relatively small, it cannot be expected that the red sequence stands out above the fore- and background distribution of galaxies. If the EXT-FOF algorithm omits parts of an obvious red sequence in the CMDs, the chosen linking parameters are too small and have to be increased.

Chapter 6

Structures in the MUNICS Survey

Having explained all the necessary ingredients for structure finding, this chapter shows the application of the EXT-FOF algorithm to the MUNICS survey. Following the recipe described here, the algorithm can be run on any other photometric redshift galaxy catalog, as well.

In Sect.6.1 an overview of the input galaxy catalog is given. Special emphasis is put on the definition of the geometric borders of the individual MUNICS fields, as well as the identification of detected objects as galaxies. Sect.6.2describes the application of the EXT-FOF algorithm to structure finding in the MUNICS survey. The utilized equations, cosmology and other parameters are listed, and the optimization of the link- ing parameters is presented. In Sect.6.3an overview is given of the entire structure catalog. The field S6F5 is used as an example to describe the quality assessment of the individual clusters and a detailed look is put on a few structures of S6F5, showing their Voronoi tessellation and likelihood maps, as well as their CMDs. Sect.6.4finally deals with the spectroscopic verification of the EXT-FOF clusters.

6.1

The galaxy catalog

MUNICS is a K0-selected photometric redshift galaxy survey, that has a 50% com- pleteness limit for point sources down to an average of m_K0 ≈19.1 mag. The survey

consists of observations in four optical passbands, B, V , R, I, and two near-infrared bands, J, and K0. The area covered by all six filters in the eight fields with the best photometric homogeneity and seeing is approximately 0.27 deg2. The survey’s most important properties are already mentioned in Sect.1.3. A detailed description of the survey motivation and layout, the photometric redshift determination, the complete- nesses of the individual fields, and the spectroscopic subset is given inDrory et al.

(2001b,2003),Snigula et al. 2002, andFeulner et al.(2003). The dataset used in this thesis was created in August 2003.

72 CHAPTER 6. STRUCTURES IN THE MUNICS SURVEY